TY - GEN
T1 - SCUT sampling and classification algorithms to identify levels of child malnutrition
AU - Baraybar-Huambo, Juan
AU - Gutiérrez-Cárdenas, Juan
N1 - Publisher Copyright:
© Springer Nature Switzerland AG 2020.
PY - 2020/1/1
Y1 - 2020/1/1
N2 - Child malnutrition results in millions of deaths every year. This condition is a potential problem in Peruvian society, especially in the rural parts of the country. The consequences of malnutrition range from physical limitations to declining mental performance and productivity for the individual. Government initiatives contribute to decreasing the causes of this disorder; however, these efforts are focused on long term solutions. The need for a fast and reliable way to detect these cases early on still exists. This paper compares classification techniques to determine which one is the most appropriate to classify cases of malnutrition. Neural networks and decision trees are used in combination with different sampling techniques, such as SCUT, SMOTE, random oversampling, random undersampling, and Tomek links. The models produced using oversampling techniques achieved high accuracies. Further, the models produced by the SCUT algorithm achieved high accuracies, preserved the behavior of the data and allowed for better representations of minority classes. The multilayer perceptron model that used the SCUT sampling techniques was chosen as the best model.
AB - Child malnutrition results in millions of deaths every year. This condition is a potential problem in Peruvian society, especially in the rural parts of the country. The consequences of malnutrition range from physical limitations to declining mental performance and productivity for the individual. Government initiatives contribute to decreasing the causes of this disorder; however, these efforts are focused on long term solutions. The need for a fast and reliable way to detect these cases early on still exists. This paper compares classification techniques to determine which one is the most appropriate to classify cases of malnutrition. Neural networks and decision trees are used in combination with different sampling techniques, such as SCUT, SMOTE, random oversampling, random undersampling, and Tomek links. The models produced using oversampling techniques achieved high accuracies. Further, the models produced by the SCUT algorithm achieved high accuracies, preserved the behavior of the data and allowed for better representations of minority classes. The multilayer perceptron model that used the SCUT sampling techniques was chosen as the best model.
KW - Child malnutrition
KW - Decision trees
KW - Neural networks
KW - Random forest
KW - Sampling techniques
UR - https://hdl.handle.net/20.500.12724/10932
UR - http://www.scopus.com/inward/record.url?scp=85084834725&partnerID=8YFLogxK
UR - https://www.mendeley.com/catalogue/4f31cddd-fdf6-3152-ab30-881aa272e343/
U2 - 10.1007/978-3-030-46140-9_19
DO - 10.1007/978-3-030-46140-9_19
M3 - Articulo (Contribución a conferencia)
AN - SCOPUS:85084834725
SN - 9783030461393
T3 - Communications in Computer and Information Science
SP - 194
EP - 206
BT - Information Management and Big Data - 6th International Conference, SIMBig 2019, Proceedings
A2 - Lossio-Ventura, Juan Antonio
A2 - Condori-Fernandez, Nelly
A2 - Valverde-Rebaza, Jorge Carlos
PB - Springer
T2 - 6th International Conference on Information Management and Big Data, SIMBig 2019
Y2 - 21 August 2019 through 23 August 2019
ER -